Dynamic re-configuration algorithms for wireless communication networks

Mobility of the wireless users creates uncertainty in demand and may lead to non-optimum use of radio network resources. Current wireless networks are reconfigured periodically by manual methods to improve network performance. These network reconfiguration and performance optimizations consume thousands of engineering hours and often are done by trial and error, resulting in less than optimum performance. The rate of network reconfiguration will increase as competitive forces throughout the world cause the service provider to change pricing structures dynamically to attract new customers, while maintaining existing customers. These changes will drastically impact the network usage. This uncertainty further reduces the period of time a non-dynamically tuned network will perform optimally. Therefore, the need to allocate the network and radio resources on the fly to maximize network infrastructure usage and minimize network costs is becoming increasingly important. This dissertation proposes a new method to allocate radio resources and evaluates the performance impact of this scheme using the configuration and operational measurements of traffic demand over a period of several days.; In the 1970s, D.C. Cox and D.O. Reudink proposed and analyzed Dynamic Channel Assignment (DCA) algorithms as a way to increase channel utilization. The algorithms attempted to maximize channel utilization under the radio channel reuse constraint by a fixed D/R. Many researchers believed that centralized-DCA approaches are extremely complex and difficult to undertake in real-time. Most DCA schemes allocate channels on an instantaneous need basis. Therefore, most of these DCA approaches lack a network-wide view of radio interference and produce limited radio performance gains.; Unlike other DCA algorithms, this research will use a centralized-DCA scheme. Due to the complexity in instantaneous DCA, we propose that an interval based DCA algorithm be used. We define a network-wide deployment model to estimate the two-way interference for candidate radio configurations. This DCA approach takes advantage of the spatial-temporal variances in the wireless networks. Each cell's channel requirements are computed based on an interval-demand estimate. Our results have demonstrated a 10-20% reduction in total allocated channels when the enhanced dynamic radio system is implemented instead of a conventional Fixed Channel Allocation (FCA). We use a two-pass approach to minimize the global interference per unit of traffic. The first pass algorithm selects radio channels which minimize potential interference. The second pass algorithm reduces the global I/C metric by reassigning radio channels. We evaluate the performance of two different channel reassignment algorithms, Cell Level Radio Tuning Algorithm (CLRTA) and Gradient Descent Algorithm (GDA). The GDA is shown to have superior performance to that of the CLRTA by approximately 2-4 dB depending on the number of available radio channels. The centralized DCA approach is approximately 3-5 dB better than the conventional FCA scheme.